Systemic Dissemination of Viral Vectors During Intratumoral Injection

Molecular Cancer Therapeutics 1233

Systemic dissemination of viral vectors during intratumoral injection

Yong Wang,1 Jim Kang Hu,2 Ava Krol,1 therapy (1, 2). However, the efficacy of gene therapy is Yong-Ping Li,2 Chuan-Yuan Li,2 and Fan Yuan1 still limited by the delivery of therapeutic genes into

1 2 target cells. Systemic delivery of viral vectors is inade- Departments of Biomedical Engineering and Radiation quate, primarily due to the poor interstitial penetration in Oncology, Duke University, Durham, NC solid tumors (3, 4) and normal tissue toxicity caused by viral vectors and/or gene products (5–15). Different approaches have been developed for reducing Abstract the toxicity in normal tissues. One is to switch to non-viral Intratumoral injection is a routine method for local viral vectors, such as cationic liposomes or polymers (16, 17). gene delivery that may improve interstitial transport of viral Non-viral vectors are less toxic and may have similar vectors in tumor tissues and reduce systemic toxicity. transfection efficiency in vitro as viral vectors. However, However, the concentration of transgene products in non-viral vectors are in general less efficient in vivo. The normal organs, such as in the liver, may still exceed normal second approach is to use tissue targeting viral vectors tissue tolerance if the products are highly toxic. The (18–23), which can be achieved through at least two elevated concentration in normal tissues is likely to be mechanisms. One is to incorporate specific molecular caused by the dissemination of viral vectors from the structures on the vector surface that can bind to unique tumor. Therefore, we investigated transgene expression in markers on the plasma membrane of cells or extracellular the liver, the serum, and a mouse mammary carcinoma matrix in tumors; another is to incorporate specific (4T1) in mice after intratumoral injection of adenoviral transcriptional promoters in viral vectors that can be vectors for mouse interleukin-12, luciferase, enhanced triggered by either endogenous factors or exogenous green fluorescence protein, or B-galactosidase. We also interventions. In all these cases, transgene expression is performed numerical simulations of virus transport in restricted in target cells or tissues. However, targeted gene tumors after intratumoral injection, based on the Krogh delivery requires identification of unique markers in cells cylinder model. Our experimental data and numerical and tissues that can capture the vectors or identification of simulations demonstrated that virus dissemination was specific transcriptional mechanisms that can control gene significant in mice and it occurred mainly during the expression. Both requirements cannot be always achieved. intratumoral injection. To reduce virus dissemination, we The two approaches mentioned above can reduce the mixed these vectors with a viscous alginate solution and toxicity in normal tissues, but they cannot solve the problem injected the mixture into the tumors. Our data showed that of poor interstitial penetration if the vectors are delivered the alginate solution could significantly reduce virus systemically. To simultaneously improve interstitial trans- dissemination while having minimal effects on transgene port and reduce normal tissue toxicity, the optimal expression in tumors and on interleukin-12-induced tumor approach is to locally inject viral vectors into tumors growth delay. These data suggest that virus dissemination (24, 25). Intratumoral injection can improve interstitial is a potential problem in local viral gene therapy of cancer transport through at least three mechanisms. One is to and that the dissemination could be significantly reduced establish a pressure gradient for enhancing convection, by the alginate solution without compromising the efficacy which is critical for delivery of macromolecules and of gene therapy. (Mol Cancer Ther. 2003;2:1233–1242) nanoparticles (3, 4). The second is to increase the pore size in tumor tissues due to pressure-induced tissue deformation (26, 27). Tissue deformation will also, as the third Introduction mechanism, improve the connectedness of interstitial space (28). When injected directly into tumors, viral vectors are Discoveries in molecular and cell biology have led to a expected to infect only cells near the injection site and hence significant development in novel strategies for cancer gene cause minimal toxicity in normal tissues. However, data in the literature have shown that the normal tissue cytotoxicity has been a limiting factor for achieving an optimal dose of Received 3/27/03; revised 7/20/03; accepted 8/7/03. viral vectors in tumors (29, 30). The concentration of The costs of publication of this article were defrayed in part by the transgene products in the liver can be on the same order payment of page charges. This article must therefore be hereby of magnitude as that in tumor tissues (31, 32). marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The normal tissue toxicity is likely to be caused by Grant support: National Science Foundation (BES-9984062) and the viruses disseminated from tumors. Virus dissemination NIH (CA81512). has been suggested in previous studies (29–34), although Requests for Reprints: Fan Yuan, Department of Biomedical Engineering, the significance of this problem and mechanisms of the 136 Hudson Hall, Box 90281, Duke University, Durham, NC 27708. Phone: (919) 660-5411; Fax: (919) 684-4488. dissemination remain to be determined. There exist at E-mail: [email protected] least two possible mechanisms of dissemination: (a) direct

injection of viral vectors into blood vessels that are Type Culture Collection, Manassas, VA), harvested at 48 h damaged by the injection needle; and (b) diffusion of after infecting cells, and purified through cesium chloride viral vectors into microvessels after intratumoral injection. gradient centrifugation according to a standard protocol To demonstrate virus dissemination, we injected adeno- (35). Viral vectors were stored in 10% glycerol at 80jC. viral vectors encoding either mouse interleukin-12 (IL-12), The virus suspensions were mixed with the solution of luciferase, or enhanced green fluorescence protein (EGFP) alginate (Sigma Chemical Co., MO) or PBS before intra- into tumors, measured IL-12 concentrations in the liver, tumoral injection. The final concentration of the alginate serum, and tumors, and compared qualitatively the was 4% w/w concentration. expressions of luciferase and EGFP in the liver and Measurement of Solution Viscosity tumors, respectively. In addition, we developed a math- Brookfield DV-III Rheometer (Brookfield, MA) was ematical model of virus transport for determining which used to measure the viscosity of solutions. The temper- of the two mechanisms mentioned above was dominant in ature of water bath was set at 37jC. The samples were virus dissemination. Our experimental data and numerical put into the center of the cup and preheated for 10 min simulations demonstrated that virus dissemination was a before the viscosity measurement. The measurement was potential problem in local viral gene therapy of cancer, repeated three times and the average value is reported in and that the dissemination occurred mainly during the this paper. intratumoral injection. EGFP Expression in Different Tissues Based on the mechanistic studies mentioned above, we Fifty microliters of AdCMVEGFP in PBS or alginate further developed a method for reducing the systemic solution were injected into 4T1 tumors via a 30-gauge dissemination of viral vectors. The method was based on needle. The dose of injection was either 1.0 108 or 5.0 the biological and physical properties of alginate solu- 108 plaque-forming units (pfu)/tumor. The samples of the tions, which are biocompatible and biodegradable and are blood, the spleen, the lung, the kidney, the heart, the liver, highly viscous. The alginate solution, when mixed with and tumors were harvested at 2 days after virus injection. viral vectors before intratumoral injection, can reduce both Then, the samples were glued onto a specimen block and the rates of convection and diffusion of viral vectors. In transferred to the stage of a Vibratome (Model 3000; addition, intratumoral injection of this solution requires a Technical Products International, St. Louis, MO) main- higher pressure gradient which may cause a transient tained at 4jC. The tissues were then sectioned into compression of blood vessels in tumors. As a result, the 300-Am slices. EGFP expression in tissues was examined virus dissemination was reduced significantly when we under a confocal laser scanning microscope (LSM 510, injected into tumors with the mixture of viral vectors and Carl Zeiss, Thornwood, NY). the alginate solution rather than the free suspension of Luciferase Expression in Different Tissues viral vectors. Fifty microliters of AdCMVLUC in PBS or alginate solution were injected into 4T1 tumors via a 30-gauge 8 Materials and Methods needle. The dose of injection was 2.0 10 pfu/tumor. At 2 days after virus injection, mice were anesthetized with Tu m o r Mo d e l s i.p. injection of 80 mg ketamine and 10 mg xylazine/kg 4T1 mouse mammary carcinoma cells were obtained body weight. Then, 50 Al of aqueous D-luciferin solution from Dr. Fred Miller’s lab (Michigan Cancer Foundation, (28.6 mg/ml) were injected into the peritoneal cavity. Detroit, MI). They were maintained in DMEM supple- Twenty minutes later, the mice were placed in the mented with 10% newborn bovine serum (Hyclone, Logan, Xenogen In Vivo Imaging System (IVIS) (Xenogen, UT) and 1% penicillin/streptomycin (Life Technologies, Alameda, CA), and a grey scale reference image was j Grand Island, NY) at 37 C with 5% CO2. One million 4T1 obtained under a low-level illumination. Then, the cells in 50 Al PBS were s.c. injected into the right hind leg of reference image was overlaid with the bioluminescence 4- to 6-week-old syngeneic Balb/c female mice (Charles image which was acquired by integrating photons River Laboratory, Wilmington, MA). The s.c. tumors were emitted from the animals for 30 s. The amount of ready for our experiments when they reached 5–8 mm bioluminescence was quantified as the number of photons in diameter. emitted per second per square centimeter per steridian Adenoviral Vectors (sr) and is shown in the pseudo-color indicated by the The Ad5-based recombinant system was used to color bar. produce the adenoviral vectors in this study. The cDNAs IL-12 Expression in Different Tissues for both mouse IL-12 subunits (p35 and 40) were inserted Fifty microliters of AdCMVIL-12 in PBS or alginate into the E1 region of the adenoviral vector, AdCMVIL-12. solution were injected into 4T1 tumors via a 30-gauge Expression of IL-12 cDNAs was driven by the cytomeg- needle. The dose of injection was 1.0 108 pfu/tumor. The alovirus (CMV) promoter. The EGFP, luciferase, and samples of the blood, the liver, and tumors were harvested h-galactosidase expressing adenoviral vectors, AdCM- at 1, 3, or 5 days after virus injection. Tumor and liver VEGFP, AdCMVLUC, and AdCMVLacZ, respectively, samples were homogenized in PBS containing the Complete were also under the control of the CMV promoter. All protease inhibitor cocktail (Boehringer Mannheim, Indian- adenoviral vectors were propagated in 293 cells (American apolis, IN). All samples were then centrifuged at 10,000 g

Table 1. Baseline values of model constantsa Statistical Analysis The Mann-Whitney U test was used to compare the R 0.01 cm S 0.27 cm2 4 11 difference between two unpaired groups. The tests were a 5.0 10 cm C0 6.15 10 M v 9 2 9 considered significant if the P values were less than 0.05. D 2.0 10 cm /s CR0 8.74 10 M v 8 5 1 Peff 2.0 10 cm/s kon 3.0 10 (Ms) Mathematical Model of Virus Dissemination after $v 4 1 0.05 koff 2.0 10 s Intratumoral Injection v 6 1 Vp 1ml kd 5.8 10 s To understand the mechanisms of virus dissemination, we developed a mathematical model of virus transport in a The determination of baseline values is discussed in the ‘‘Materials and Methods’’ section. tumor tissues after intratumoral injection. The model was based on the Krogh cylinder geometry and considered the interstitial diffusion of viruses, binding of viruses to their receptors on the cell membrane, internalization of viruses for 10 min. The supernatants were extracted and stored at – into cells, and inactivation of viruses in tumors. In 80jC. IL-12 concentration in different supernatants was general, convection should also be considered in transport quantified by an ELISA kit with a detection sensitivity of 2.5 analysis of viruses. However, convection was negligible in pg/ml (R&D Systems, Minneapolis, MN). The results were solid tumors (3). The diffusion in this model was assumed then converted to the concentrations of IL-12 in different to be time-dependent and only in the radial direction, and tissues based on the sample volume or weight. the rate of virus inactivation was assumed to be BB-Galactosidase Expression in Tumor Tissues proportional to the local concentrations of viruses. With Fifty microliters of AdCMVLacZ in PBS or alginate these considerations and assumptions, the governing solution were injected into 4T1 tumors via a 30-gauge equation for free virus transport was derived as 8 needle. The dose of injection was 2.0 10 pfu/tumor. Tumor samples were harvested at 2 days after virus @Cv Dv @ @Cv i ¼ r i k CvCv =$v þ k Cv kvCv ðAÞ injection. Tissue samples were fixed in 4% paraformalde- @t r @r @r on i R off B d i hyde in PBS for 24 h and rinsed with 100 mM sodium- phosphate buffer (pH 7.3, 2 mM MgCl2, 0.1% Triton X-100) v v v where Ci , CB, CR were the concentrations of free viruses, twice for 1 h. Tissues were then mounted on the Vibratome bound viruses, and available virus receptors in tumors, A and sectioned into 300- m slices. The tissue slices were then respectively; t was time; r was the radial distance; Dv was incubated at 25jC overnight in a staining solution with v the diffusion coefficient of viruses; kd was the rate 1 mg/ml X-gal (Life Technologies), 5 mM potassium constant of free virus inactivation; and $$v was the ferricyanide, 5 mM potassium ferrocyanide, 2 mM MgCl2, available volume fraction of viruses in tumors. The overall and 0.1% Triton X-100. The stained slices were scanned into association and dissociation rate constants were approx- a computer, using a Plustek Optic Pro document scanner imately equal to the intrinsic rate constants, kon and koff, (Model 12000P). v respectively, because 4kacD >> kon if ac was the radius Tumor Size Measurement v of a cell (36) (see also Table 1). CR was related with the Tumor size was measured at different time points after total concentration of virus receptor, CR0,by virus injection. A caliper was used to measure the longest v v (L) and the shortest (W) dimensions of the tumors. The CR ¼ CR0 CB ðBÞ tumor was assumed as an ellipsoid. Thus, its volume 2 could be calculated as: Vt =(p/6)LW . The relative The governing equation for bound viruses was derived as volume of tumors, Vr, was defined as: Vr = Vt/V0, where @ v Vt was measured at different time points after virus CB v v $v v v v v ¼ konCi CR= koff CB kdCB Rin ðCÞ injection and V0 was the tumor volume on the day of @t virus injection. Tr a n s f e c t i o n o f C e ll s in Vitro where we assumed that the rate constant of bound virus Alginate was dissolved in cell culture medium at inactivation was the same as that of free viruses. concentrations of 0.1% and 4% w/w, respectively. 4T1 cells Furthermore, we assumed that the rate of internalization R v were cultured in the 24-well plates until they reached of bound viruses, in, was much smaller than the rate of k vC v R v 80–90% confluence. Two hundred-microliter virus suspen- inactivation, d B. Thus, in in Eq. C was neglected in sions in the culture medium or alginate solutions were first our numerical simulations. added into the wells. The number ratio of viral particles The amount of intravasated viruses per unit volume v versus cells was on the order of 10. After virus suspension of the plasma, Cp, could be calculated as was placed in the well, 600 Al culture medium was slowly dCv Pv S added into the well to avoid the mixing with the virus p eff v $v ¼ Ci = ðDÞ suspension. After 48 h, the cells were examined under a dt Vp fluorescence microscope. The same experiment was repeated three times in different days and the typical results are where we assumed that the concentration of viruses in the reported in the ‘‘Results’’ section. plasma was much smaller than that in the tumor. This

assumption was verified later by the numerical simula- (see Table 1), based on the size of viruses and the data of v tions as shown in Fig. 5. In Eq. D, Peff was the effective macromolecules and liposomes in tumor tissues (37, 41– microvascular permeability of viruses, S was the total 44). The intrinsic rate constants of association and surface area of tumor vessels exposed to viruses, and Vp dissociation between adenoviruses and their receptors, 5 4 1 was the plasma volume in the body. kon and koff, were 3.0 10 (Ms) 1 and 2.0 10 s , The initial conditions were: respectively (45). Our experimental data showed that the peak of IL-12 concentration in mice was reached in less v Ci ¼ C0 ðE1Þ than 1 day after virus injection. Thus, the half-life of adenoviruses in mice was less than 1 day. Based on this information, we assumed that the rate constant of virus v v 1 6 1 CB ¼ 0 ðE2Þ inactivation, kd, was 0.5 day or 5.8 10 s . Results Cv ¼ 0 ðE3Þ p Dissemination of Adenoviral Vectors for EGFP and Luciferase C where 0 was the initial virus concentration in the tumor. It EGFP is an exogenous and non-secretable protein. It was assumed to be equal to the amount of viruses left in the should not be detected in normal tissues after intratumoral tumor immediately after intratumoral injection divided by injection of AdCMVEGFP unless the viral vectors escaped the distribution volume of virus in tumors. The boundary from tumors. In our experiment, we observed EGFP conditions were expression in both liver (Fig. 1A) and tumors (Fig. 1B) at 2 days after intratumoral injection of 5.0 108 pfu @Cv i ¼ 0 at r ¼ R ðF1Þ AdCMVEGFP. These results indicated that AdCMVEGFP @r could disseminate into the liver. EGFP expression in the liver could not be detected when the dose of injection was 8 v reduced to 1.0 10 pfu/tumor. @C v Dv i ¼ Pv Cv=$ at r ¼ a ðF2Þ To reduce AdCMVEGFP dissemination, we mixed the @r eff i vectors with the alginate solution before the injection. The where R was the radius of the Krogh cylinder and a was final concentration of alginate in the mixture was 4%. the radius of blood vessel. The viscosity of pure alginate solution at the same BaselineValues of Model Constants concentration was 1130 cp, which was three orders of magnitude higher than that of PBS. Mixing viral vectors The baseline values of model constants are shown in with the alginate solution significantly reduced EGFP Table 1. We assumed that the radii of the microvessel, a, expression in the liver after intratumoral injection of 5.0 and the Krogh cylinder, R, were 5 and 100 Am, respective- 108 pfu AdCMVEGFP (Fig. 1C), but had minimal effects ly, and the plasma volume, V , in a mouse was 1 ml. The p on EGFP expression in tumors (Fig. 1D). These results distribution volume of viruses in tumors, V , could be d demonstrated that virus dissemination could be signifi- approximated by V f 3/(1 e ), where V was the inj c inj cantly reduced by the alginate solution. volume of injection, f was the retardation coefficient of We could not detect EGFP, under a regular fluorescence convective transport (26), and e was the volume fraction c microscope, in tissue samples from the serum, the spleen, of cells. In our experiment, V =50Al. We assumed that inj the lung, the kidney, and the heart after intratumoral f = 0.3 and e = 0.5 (37). Thus, V =2.7Al. Within this c d injection of 5.0 108 pfu AdCMVEGFP. These results were volume, the total surface area of tumor microvessels, S, consistent with the data in the literature, indicating that the could be calculated as e V (S/V), where e was the ves d ves liver was the main target organ of disseminated viruses volume fraction of tumor vessels and S/V was the ratio of (29, 31, 32). the surface area versus the volume of tumor microvessels. 2 3 We also injected AdCMVLUC into tumors. The pattern We assumed that eves = 5% (38) and S/V = 2000 cm /cm (39). Thus, S = 0.27 cm2. The initial concentration of the of luciferase expression was similar to that of EGFP (Fig. 2A). It was strong in the tumor and the liver, but viruses, C0, was calculated as the number of viruses could not be detected in other organs. When the viral divided by Vd and the Avogadro’s number. The number of viruses in tumors immediately after injection was assumed vectors were mixed with the alginate solution, luciferase 8 11 expression in the liver was significantly reduced (Fig. 2B). to be 10 pfu. Thus, C0 = 6.15 10 M. The number of adenovirus receptors is approximately 10,000 per cell (40) These results confirmed the observations, shown in Fig. 1, and the volume of a cell is approximately 950 Am3 (37). in the EGFP experiment. Dissemination of an Adenoviral Vector for IL-12 Thus, the total concentration of the receptor, CR0, was 9 v assumed to be 8.74 10 M. The diffusion coefficient, D , Results shown in Figs. 1 and 2 are qualitative, since we v the effective microvascular permeability, Peff, and the could not quantify the concentration of EGFP and luciferase available volume fractions, $v, of adenoviruses in tumors in tissues. To quantify transgene expression, we investi- are still unknown. Their values were assumed in our study gated the dissemination of an adenoviral vector for IL-12,

Figure 1. Comparison of EGFP expression in the livers (A, C) and the tumors (B, D) after intratumoral injection of AdCMVEGFP at the dose of 5.0 108 pfu/tumor. In A and B, the fluorescence images show EGFP distributions in the liver and the tumor, respectively, treated with AdCMVEGFP in PBS. In C and D, the fluorescence images show EGFP distributions in the liver and the tumor, respectively, treated with AdCMVEGFP in the alginate solution. Scale bars, 100 Am.

AdCMVIL-12. In this experiment, we injected AdCMVIL-12 Again, the reduction was statistically significant (P < 0.05). into 4T1 tumors at the dose of 1.0 108 pfu/tumor and The serum concentrations of IL-12 in both groups quantified IL-12 concentrations in the serum, liver, and decreased with time; and the differences between the two tumors after intratumoral injection. IL-12 expression in groups could not be detected on days 3 and 5. On day 5, other tissues was not determined, since the liver was the the concentration of IL-12 ml in the serum was less than main target organ of the disseminated viruses as discussed 0.1 ng/g in both groups. above. The concentration of IL-12 was time-dependent The time dependence of IL-12 concentration in tumor (Fig. 3). The peak concentrations of IL-12 in all tissues were tissues was similar to those in the liver and serum, reached on day 1 post-intratumoral injection (Fig. 3). respectively (Fig. 3). However, there was no significant The concentration of IL-12 in the liver was reduced difference in the IL-12 concentration between PBS and approximately 8-fold, if the viral vectors were mixed with alginate groups at all time points (Fig. 3C) (P > 0.05). These thealginatesolutionbeforeinjection(Fig.3A).The results indicated that the net effect of alginate solution on reduction was statistically significant (P < 0.05). On days transgene expression in tumors was negligible. Possible 3 and 5, IL-12 concentration in the liver was between 0.3 explanations of these results will be discussed later. and 0.4 ng/g, and there was no significant difference in the IL-12 is a cytokine that has a potential to treat human IL-12 concentrations between PBS and alginate groups infectious and malignant diseases. Thus, we also quantified (Fig. 3A). the effects of AdCMVIL-12 treatment on tumor growth. We The concentration profile of IL-12 in the serum was observed that the growth of tumors treated with AdCM- similar to that in the liver and consistent with the results VIL-12 was slower than that treated with AdCMVEGFP in a previous report (33). The peak concentration was (Fig. 4). The difference in tumor size was statistically reduced approximately 4-fold, if the viral vectors were significant from day 6 to day 21 after intratumoral injection mixed with the alginate solution before injection (Fig. 3B). of viral vectors. Quantitatively, the tumor doubling time

Figure 2. Comparison of luciferase expression in the livers and the tumors treated with AdCMV- LUC in PBS (A) and the alginate solution (B). The dose of AdCMVLUC injection was 2.0 108 pfu/ tumor. The amount of bioluminescence, which is related to the concentration of luciferase, was quantified as the number of photons emitted per second per square centimeter per steridian (sr) and is shown in pseudo-color indicated by the color bar.

was increased from 3 to 6 days in the control group to The concentrations of bound and intravasated viruses 18–21 days in AdCMVIL-12-treated groups (Fig. 4). Mixing shown in Fig. 5B depend also on the concentration and the AdCMVIL-12 with the alginate solution had minimal distribution of virus receptors on the cell membrane; and effects on the efficacy of AdCMVIL-12 treatment (P > 0.05) the receptor distribution in a tumor can be heterogeneous. (Fig. 4), presumably because it did not change IL-12 To understand the effects of receptor concentration and concentration in tumors (see Fig. 3C). We also quantified distribution on virus dissemination, we varied the receptor IL-12 concentration in tumors at 1 day after AdCMVEGFP concentration by three orders of magnitude in our injection, and observed that it was three orders of numerical simulations. The simulation results demonstrat- magnitude lower than that in AdCMVIL-12-treated tumors. ed that the amount of bound viruses was decreased and Virus Intravasation after Intratumoral Injection the amount of intravasated viruses was increased when the Results shown in Figs. 1–3 demonstrated clearly that receptor concentration was decreased. However, the viral vectors could disseminate into the systemic circulation amount of intravasated viruses was always much less than during and/or after intratumoral injection. To understand that of bound viruses. Therefore, the heterogeneous the mechanisms of virus dissemination, we performed distribution of virus receptors within a tumor or among numerical simulations of virus diffusion into microvessels different tumors should have minimal effects on virus (i.e., the intravasation process), based on the Krogh cylinder dissemination. model. Results from numerical simulations indicated that Effects of Alginate Solution onVirus Distribution and the concentration of free viruses in the tumor decreased Transfection Efficiency rapidly with time and became negligible, compared with To investigate the effects of alginate solution on virus that of bound viruses, during the first few hours after the distribution and its ability to transfect tumor cells, we virus injection (data not shown). The concentration of performed two additional experiments. In the first exper- bound viruses decreased gradually with time, except that iment, we divided tumors into three groups. In the first and near the microvessel wall (Fig. 5A). In this region, the second groups, adenoviral vectors for h-galactosidase in spatial gradient of the concentration of bound viruses was PBS and alginate solution, respectively, were injected into also large (Fig. 5A). When the average concentration of tumors. In the third group, we injected Evans Blue dye into bound viruses in tumors was calculated, we observed that tumors. At 48 h after adenovirus injection or immedi- it decreased gradually with time (Fig. 5B). The amount of ately after Evans Blue injection, we observed that both intravasated viruses per unit volume of the plasma h-galactosidase and Evans Blue dye accumulated at the increased gradually with time (Fig. 5B). Quantitatively, periphery of tumors in all three groups (images not shown). the data shown in Fig. 5B indicated that less than one viral The similarity in these spatial distributions suggested that particle could escape from the tumor during a 3-day the alginate solution had insignificant effects on viral vector period. Therefore, virus dissemination after intratumoral distribution in tumor tissues, and that the distribution of injection was negligible; it could not explain the transgene transgene expression was determined mainly by the expression in the liver shown in Figs. 1–3. distribution of viral vectors immediately after injection.

since the alginate solution at the concentration of 0.1% had no significant effects on the number of transfected cells (Fig. 6C). These results suggested that the alginate solution could reduce the rate of virus transport in solutions near the surface of cells, and thus decrease the number of viral vectors that could infect tumor cells.

Discussion We investigated systemic dissemination of viral vectors in local viral gene delivery. Our experimental data showed that the viral vectors injected into tumors could disseminate into normal tissues and the amount of dissemination was significant. Meanwhile, our numerical simulations indicated that interstitial diffusion of viral vectors after intratumoral injection could not account for the amount of disseminated viral vectors observed in our experiments. Therefore, the virus dissemination might occur during the intratumoral injection. In the second part of the study, we developed a method to reduce the virus dissemination through mixing the virus suspension with the alginate solution before the injection. The alginate solution significantly reduced the virus dissemination but had minimal effects on the concentration of transgene products in tumors. Two issues in our study need to be discussed to understand these experimental data. One is the mechanisms of the alginate-mediated reduction in virus dissemination. Another is why the reduction in virus dissemination did not lead to an increase in the transgene expression in tumors. The mechanisms of the reduction in virus dissemination might be 2-fold. As discussed above, the virus dissemination occurred during the intratumoral injection. More

Figure 3. IL-12 concentrations in the liver (A), the serum (B), and the tumor (C) at different time points after intratumoral injection of AdCMVIL- 12 in the alginate solution or PBS. The dose of injection was 1.0 108 pfu/ tumor. Error bars, SE in four tumors. The asterisk indicates that the difference in IL-12 concentrations between alginate and PBS groups is statistically significant (P < 0.05).

In the second experiment, we treated 4T1 cells cultured in 24-well plates with the adenoviral vector for EGFP for 48 h. The vectors were suspended in either PBS or alginate Figure 4. Tumor growth in mice treated with either AdCMVEGFP in the alginate solution (i.e., the control), AdCMVIL-12 in PBS, or AdCMVIL-12 in solutions with two different concentrations: 0.1% and 4%. the alginate solution. The data are reported as the volume of tumors At the end of treatment, we observed that the alginate relative to that in the same animals immediately before the treatment. The solution at the concentration of 4% reduced the number of injection doses of AdCMVEGFP and AdCMVIL-12 were all equal to 1.0 8 transfected cells (Fig. 6A), compared with that in the PBS 10 pfu/tumor. Symbols, mean of the relative volume of six tumors; error bars, SE. The asterisks indicate that the differences in size between group (Fig. 6B). The reduction was unlikely to be due to AdCMVIL-12- and AdCMVEGFP-treated tumors at the same time points alginate-induced changes in the bioactivity of viral vectors are statistically significant (P < 0.05).

viscosity of the alginate solution was one to two orders of magnitude higher than that of the blood. Therefore, both the compression and the sealing of damaged vessels could be potential mechanisms of the alginate-mediated reduction in virus dissemination. To address the second issue, we need to consider three possible scenarios. First, alginate molecules might react chemically with the viral particles. The reaction would reduce the bioactivity of viral vectors in tumors. However, we argued that this possibility was very small because alginate is a biocompatible and biodegradable polysaccha- ride. It has been used for three-dimensional cell culture, tissue engineering, and virus encapsulation (46–51). Alginate does not affect the infectivity or bioactivity of viruses based on the plaque assay, even though the viruses have been heated and vortexed during the immobilization

Figure 5. Numerical simulations of virus transport after intratumoral injection: (A) spatial and temporal distribution of bound virus concentra- 14 tion in tumors, C t,(10 mol/ml) and (B) spatial average of the 14 concentration of bound viruses in tumors, (C t)a, (10 mol/ml) and the 25 amount of intravasated viruses per unit volume of the plasma, C p, (10 mol/ml). The time indicates the period after intratumoral injection of viral vectors and the distance indicates the location from the axis of the blood vessel in the Krogh cylinder model.

specifically, the dissemination could result from a direct injection of the viral vectors into the blood vessels that were damaged by the injection needle. When the viral vectors were mixed with the alginate solution, the blood vessels near the injection site could be compressed by the high interstitial fluid pressure caused by the injection. This was because the viscosity of the alginate solution was three orders of magnitude higher than that of PBS. For injecting the same amount of viral suspension over the same time period, we had to apply a much higher pressure for the alginate solution than for the PBS. When the blood vessels were compressed, the pathways for virus dissemination were transiently blocked. After virus injection, the interstitial fluid pressure was reduced but the damaged vessels could still be sealed by the alginate solution. In fact, we had observed small pieces of alginate gels near the injection site in tumor tissues. Even if a small amount of alginate solution was injected into the blood Figure 6. Comparison of EGFP expression in cultured 4T1 cells treated with AdCMVEGFP for 48 h in 4% alginate solution (A), cell culture vessels, the flow of the solution was negligible under a medium (B), and 0.1% alginate solution (C). The images of cells with normal blood pressure gradient. This was because the EGFP were acquired at the end of the treatment.

and lyophilization processes (50, 51). In our study, viral therapy and that virus dissemination can be reduced vectors were only gently mixed with the alginate solution significantly by the alginate solution without compromis- at the room temperature immediately before the intra- ing the therapeutic efficacy of the therapy. This approach tumoral injection. Thus, the bioactivity of viral vectors in is not limited to the delivery of adenoviral vectors; it may our experiments was unlikely to be reduced by alginate be more useful for intratumoral injection of highly potent molecules. This conclusion was also supported by the in and self-replicating viral vectors, because systemic toxic- vitro data shown in Fig. 6 when the concentration of ity is higher for these vectors once they leak out into alginate was low (0.1% w/w). normal tissues. Second, alginate molecules might interact physically with the viral particles. The interaction would significant- Acknowledgments ly decrease the diffusion coefficient of viral particles since We thank Dr. Qing Chen for his technical assistance and Drs. David Katz the diffusion coefficient is inversely proportional to the and Marcus Henderson for their help in viscosity measurement. viscosity of alginate solutions. Data shown in Fig. 6 were consistent with this scenario. However, the amount of viral particles that can diffuse from the middle of a References solution to the surface of tumor cells over a fixed time 1. Verma, I. M. and Somia, N. Gene therapy—promises, problems and prospects. Nature, 389: 239 – 342, 1997. period is also proportional to the square of the diffusion 2. Anderson, W. F. Human gene therapy. Nature, 392: 25 – 30, 1998. distance which was very small (< 1 Am) in porous tumor f 3. Jain, R. K. Delivery of molecular and cellular medicine to solid tumors. tissues but large ( 1 mm) in cell culture medium. Microcirculation, 4: 1 – 23, 1997. Therefore, the total amount of viral particles that could 4. McGuire, S., Zaharoff, D., and Yuan, F. Interstitial transport of infect tumor cells in vivo was less affected by the alginate macromolecules: implication for nucleic acid delivery in solid tumors. In: R. I. Mahato and S. W. Kim (eds.), Pharmaceutical Perspectives of Nucleic solution in the experiments than that in vitro. On the other Acid-Based Therapeutics, pp. 434 – 454. London: Taylor & Francis Books, hand, alginate could trap viral vectors since it could exist 2002. as a gel in the tumor. The trapped viral particles could not 5. Crystal, R. G. Transfer of genes to humans: early lessons and obstacles reach tumor cells. Taken together, the alginate-mediated to success. Science, 270: 404 – 410, 1995. reduction in virus transport might explain partly why the 6. Ledley, F. D. Pharmaceutical approach to somatic gene therapy. Pharm. Res., 13: 1595 – 1614, 1996. transgene expression in tumors was not increased when 7. Lozier, J. N., Metzger, M. E., Donahue, R. E., and Morgan, R. A. the virus dissemination was decreased by the alginate Adenovirus-mediated expression of human coagulation factor IX in the solution. rhesus macaque is associated with dose-limiting toxicity. Blood, 94: 3968 – 3975, 1999. Third, it can be shown mathematically that the percent- 8. Lehrman, S. Virus treatment questioned after gene therapy death. age of alginate-mediated reduction in virus dissemination Nature, 401: 517 – 518, 1999. is always larger than the percent increase in the number of 9. Marshall, E. Gene therapy death prompts review of adenovirus vector. viral particles retained in the tumor if the total amount of Science, 286: 2244 – 2245, 1999. disseminated viruses is less than 50% of the injected dose. 10. Stone, D., David, A., Bolognani, F., Lowenstein, P. R., and Castro, For example, we may consider a case in which mixing the M. G. Viral vectors for gene delivery and gene therapy within the endocrine system. J. Endocrinol., 164: 103 – 118, 2000. viruses with the alginate solution leads to a reduction in 11. Brand, K., Arnold, W., Bartels, T., Lieber, A., Kay, M. A., Strauss, M., virus dissemination by 8-fold (see Fig. 3A). In this case, the and Dorken, B. Liver-associated toxicity of the HSV-tk/GCV approach and virus concentration in the tumor is increased by only 58% if adenoviral vectors. Cancer Gene Ther., 4: 9 – 16, 1997. we assume the amount of disseminated viruses in PBS to be 12. van der Eb, M. M., Cramer, S. J., Vergouwe, Y., Schagen, F. H., van Krieken, J. H., van der Eb, A. J., Rinkes, I. H., van de Velde, C. J., 40% of the injected dose. If the amount of disseminated and Hoeben, R. C. Severe hepatic dysfunction after adenovirus-mediated viruses in PBS is 20% of the injected dose, then mixing the transfer of the herpes simplex virus thymidine kinase gene and ganciclovir viruses with the alginate solution only results in an increase administration. Gene Ther., 5: 451 – 458, 1998. in virus concentration in the tumor by 23%. This analysis 13. Bustos, M., Sangro, B., Alzuguren, P., Gil, A. G., Ruiz, J., Beraza, N., Qian, C., Garcia-Pardo, A., and Prieto, J. Liver damage using suicide genes. suggests that the alginate solution is more effective in A model for oval cell activation. Am. J. Pathol., 157: 549 – 559, 2000. reducing virus dissemination than increasing virus con- 14. Siegel, J. P. and Puri, R. K. Interleukin-2 toxicity. J. Clin. Oncol., 9: centration in the tumor. 694 – 704, 1991. Taken together, the reduced transport of viral vectors 15. Golab, J. and Zagozdzon, R. Antitumor effects of interleukin-12 in pre- and the way of calculation were potential mechanisms for clinical and early clinical studies. Int. J. Mol. Med., 3: 537 – 544, 1999. 16. Luo, D. and Saltzman, W. M. Synthetic DNA delivery systems. Nat. explaining why there was no increase in the transgene Biotechnol., 18: 33 – 37, 2000. expression in tumors when the virus dissemination was 17. Putnam, D., Gentry, C. A., Pack, D. W., and Langer, R. Polymer-based reduced significantly by the alginate solution. These effects gene delivery with low cytotoxicity by a unique balance of side-chain of the alginate solution could not be achieved through termini. Proc. Natl. Acad. Sci. USA, 98: 1200 – 1205, 2001. decreasing the dose of injection, because the dose decrease 18. Datta, R., Rubin, E., Sukhatme, V., Qureshi, S., Hallahan, D., Weichselbaum, R. R., and Kufe, D. W. Ionizing radiation activates would proportionally reduce the total number of viral transcription of the EGR1 gene via CArG elements. Proc. Natl. Acad. particles in tumors. Consequently, transgene expression in Sci. USA, 89: 10149 – 10153, 1992. tumors would be decreased as well. 19. Weichselbaum, R. R., Hallahan, D. E., Beckett, M. A., Mauceri, H. J., Lee, H., Sukhatme, V. P., and Kufe, D. W. Gene therapy targeted by In summary, data in this study suggest that virus radiation preferentially radiosensitizes tumor cells. Cancer Res., 54: dissemination is a potential problem in local viral gene 4266 – 4269, 1994.

20. Richards, C. A., Austin, E. A., and Huber, B. E. Transcriptional M. W., and Li, C. Enhancement of radiotherapy by hyperthermia- regulatory sequences of carcinoembryonic antigen: identification and use regulated gene therapy. Int. J. Radiat. Oncol. Biol. Phys., 48: 1513 – with cytosine deaminase for tumor-specific gene therapy. Hum. Gene 1518, 2000. Ther., 6: 881 – 893, 1995. 35. Graham, F. L. and Prevec, L. Manipulation of adenovirus vectors. In: 21. Spitzweg, C., Zhang, S., Bergert, E. R., Castro, M. R., McIver, B., E. J. Murray (ed.), Gene Transfer and Expression Protocols, pp. 109 – 128. Heufelder, A. E., Tindall, D. J., Young, C. Y., and Morris, J. C. Prostate- Clifton, NJ: The Humana Press Inc., 1991. specific antigen (PSA) promoter-driven androgen-inducible expression of 36. DeLisi, C. The biophysics of ligand-receptor interactions. Q. Rev. sodium iodide symporter in prostate cancer cell lines. Cancer Res., 59: Biophys., 13: 201 – 230, 1980. 2136 – 2141, 1999. 37. Krol, A., Maresca, J., Dewhirst, M. W., and Yuan, F. Available volume 22. Huang, Q., Hu, J. K., Lohr, F., Zhang, L., Braun, R., Lanzen, J., Little, fraction of macromolecules in a fibrosarcoma: implications for drug J. B., Dewhirst, M. W., and Li, C. Y. Heat-induced gene expression as a delivery. Cancer Res., 59: 4136 – 4141, 1999. novel targeted cancer gene therapy strategy. Cancer Res., 60: 3435 – 3439, 2000. 38. Jain, R. K. Transport of molecules in the tumor interstitium: a review. Cancer Res., 47: 3039 – 3051, 1987. 23. Yamamoto, M., Alemany, R., Adachi, Y., Grizzle, W. E., and Curiel, D. T. Characterization of the cyclooxygenase-2 promoter in an adenoviral 39. Yuan, F., Baxter, L. T., and Jain, R. K. Pharmacokinetic analysis in vector and its application for the mitigation of toxicity in suicide gene two-step approaches using bifunctional and enzyme-conjugated anti- therapy of gastrointestinal cancers. Mol. Ther., 3: 385 – 394, 2001. bodies. Cancer Res., 51: 3119 – 3130, 1991. 24. Lu, Y., Carraher, J., Zhang, Y., Armstrong, J., Lerner, J., Rogers, 40. Wickham, T. J., Shuler, M. L., Hammer, D. A., Granados, R. R., and W. P., and Steiner, M. S. Delivery of adenoviral vectors to the prostate Wood, H. A. Equilibrium and kinetic analysis of Autographa californica for gene therapy. Cancer Gene Ther., 6: 64 – 72, 1999. nuclear polyhedrosis virus attachment to different insect cell lines. J. Gen. Virol., 73: 3185 – 3194, 1992. 25. Paielli, D. L., Wing, M. S., Rogulski, K. R., Gilbert, J. D., Kolozsvary, A., Kim, J. H., Hughes, J., Schnell, M., Thompson, T., and Freytag, S. O. 41. Berk, D. A., Yuan, F., Leunig, M., and Jain, R. K. Direct in vivo Evaluation of the biodistribution, persistence, toxicity, and potential of measurement of targeted binding in a human tumor xenograft. Proc. Natl. germ-line transmission of a replication-competent human adenovirus Acad. Sci. USA, 94: 1785 – 1790, 1997. following intraprostatic administration in the mouse. Mol. Ther., 1: 42. Pluen, A., Boucher, Y., Ramanujan, S., McKee, T. D., Gohongi, T., 263 – 274, 2000. di Tomaso, E., Brown, E. B., Izumi, Y., Campbell, R. B., Berk, D. A., and 26. McGuire, S. and Yuan, F. Quantitative analysis of intratumoral Jain, R. K. Role of tumor-host interactions in interstitial diffusion of infusion of color molecules. Am. J. Physiol., 281: H715 – H721, 2001. macromolecules: cranial vs. subcutaneous tumors. Proc. Natl. Acad. Sci. USA, 98: 4628 – 4633, 2001. 27. Zhang, X-Y., Luck, J., Dewhirst, M. W., and Yuan, F. Interstitial hydraulic conductivity in a fibrosarcoma. Am. J. Physiol., 279: H2726 – 43. Yuan, F., Leunig, M., Huang, S. K., Berk, D. A., Papahadjopoulos, D., H2734, 2000. and Jain, R. K. Microvascular permeability and interstitial penetration of sterically stabilized (Stealth) liposomes in a human tumor xenograft. 28. Yuan, F., Krol, A., and Tong, S. Available space and extracellular Cancer Res., 54: 3352 – 3356, 1994. transport of macromolecules: effects of pore size and connectedness. Ann. Biomed. Eng., 29: 1150 – 1158, 2001. 44. Yuan, F. Transvascular drug delivery in solid tumors. Semin. Radiat. Oncol., 8: 164 – 175, 1998. 29. Toloza, E. M., Hunt, K., Swisher, S., McBride, W., Lau, R., Pang, S., Rhoades, K., Drake, T., Belldegrun, A., Glaspy, J., and Economou, J. S. In 45. Lortat-Jacob, H., Chouin, E., Cusack, S., and van Raaij, M. J. Kinetic vivo cancer gene therapy with a recombinant interleukin-2 adenovirus analysis of adenovirus fiber binding to its receptor reveals an avidity vector. Cancer Gene Ther., 3: 11 – 17, 1996. mechanism for trimeric receptor-ligand interactions. J. Biol. Chem., 276: 9009 – 9015, 2001. 30. Tjuvajev, J. G., Chen, S. H., Joshi, A., Joshi, R., Guo, Z. S., Balatoni, J., Ballon, D., Koutcher, J., Finn, R., Woo, S. L., and Blasberg, R. G. 46. Smidsrod, O. and Skjak-Braek, G. Alginate as immobilization matrix Imaging adenoviral-mediated herpes virus thymidine kinase gene transfer for cells. Trends Biotechnol., 8: 71 – 78, 1990. and expression in vivo. Cancer Res., 59: 5186 – 5193, 1999. 47. Klock, G., Frank, H., Houben, R., Zekorn, T., Horcher, A., Siebers, U., 31. Bramson, J. L., Hitt, M., Gauldie, J., and Graham, F. L. Pre-existing Wohrle, M., Federlin, K., and Zimmermann, U. Production of purified immunity to adenovirus does not prevent tumor regression following alginates suitable for use in immunoisolated transplantation. Appl. Micro- intratumoral administration of a vector expressing IL-12 but inhibits virus biol. Biotechnol., 40: 638 – 643, 1994. dissemination. Gene Ther., 4: 1069 – 1076, 1997. 48. Becker, T. A., Kipke, D. R., and Brandon, T. Calcium alginate gel: a 32. Zhang, R., Straus, F. H., and DeGroot, L. J. Effective genetic therapy biocompatible and mechanically stable polymer for endovascular emboli- of established medullary thyroid carcinomas with murine interleukin-2: zation. J. Biomed. Mater. Res., 54: 76 – 86, 2001. dissemination and cytotoxicity studies in a rat tumor model. Endocrinol- 49. Langer, R. and Vacanti, J. P. Tissue engineering. Science, 260: 920 – ogy, 140: 2152 – 2158, 1999. 926, 1993. 33. Nasu, Y., Bangma, C. H., Hull, G. W., Lee, H. M., Hu, J., Wang, J., 50. Offit, P. A., Khoury, C. A., Moser, C. A., Clark, H. F., Kim, J. E., and McCurdy, M. A., Shimura, S., Yang, G., Timme, T. L., and Thompson, Speaker, T. J. Enhancement of rotavirus immunogenicity by microencap- T. C. Adenovirus-mediated interleukin-12 gene therapy for prostate sulation. Virology, 203: 134 – 143, 1994. cancer: suppression of orthotopic tumor growth and pre-established lung 51. Kalyanasundaram, S., Feinstein, S., Nicholson, J. P., Leong, K. W., metastases in an orthotopic model. Gene Ther., 338 – 349, 1999. 6: and Garver, R. I., Jr. Coacervate microspheres as carriers of recombinant 34. Lohr, F., Hu, K., Huang, Q., Zhang, L., Samulski, T. V., Dewhirst, adenoviruses. Cancer Gene Ther., 6: 107 – 112, 1999.

Yong Wang, Jim Kang Hu, Ava Krol, et al.

Mol Cancer Ther 2003;2:1233-1242.

Updated version Access the most recent version of this article at: http://mct.aacrjournals.org/content/2/11/1233

Cited articles This article cites 49 articles, 19 of which you can access for free at: http://mct.aacrjournals.org/content/2/11/1233.full#ref-list-1

Citing articles This article has been cited by 8 HighWire-hosted articles. Access the articles at: http://mct.aacrjournals.org/content/2/11/1233.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mct.aacrjournals.org/content/2/11/1233. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.